DISCUSSION PAPER
April 2013

RFF DP 13-10
Mercury and Air
Toxics Standards
Analysis
Deconstructed
Changing Assumptions,
Changing Results
B l ai r Be asl e y, M att W oerm an, An t hon y P a ul ,
Dallas Burtraw , and Karen Palmer
1616 P St. NW
Washington, DC 20036
202-328-5000 www.rff.org
Mercury and Air Toxics Standards Analysis Deconstructed:
Changing Assumptions, Changing Results
Blair Beasley, Matt Woerman, Anthony Paul, Dallas Burtraw, and Karen Palmer
Abstract
Several recent studies have used simulation models to quantify the potential effects of recent
environmental regulations on power plants, including the Mercury and Air Toxics Standards (MATS),
one of the US Environmental Protection Agency’s most expensive regulations. These studies have
produced inconsistent results about the effects on the industry, making general conclusions difficult. We
attempt to reconcile these differences by representing the variety of assumptions in these studies within a
common modeling platform. We find that the assumptions, and their differences from the way MATS will
be implemented, make a substantial impact on projected retirement of coal-fired capacity and generation,
investments that are required, and emissions reductions. Almost uniformly, the actual regulation, when
examined in its final form and in isolation, provides more flexibility than is represented in most models.
We find this leads to a smaller impact on the composition of the electricity generating fleet than most
studies have predicted.
Key Words: sulfur dioxide, mercury, air toxics, nitrogen oxides, carbon dioxide, electricity,
technology, generation
JEL Classification Numbers: Q47, Q53, Q58
© 2013 Resources for the Future. All rights reserved. No portion of this paper may be reproduced without
permission of the authors.
Discussion papers are research materials circulated by their authors for purposes of information and discussion.
They have not necessarily undergone formal peer review.
Contents
1. Introduction ......................................................................................................................... 1
2. EPA Policies That Are Modeled ...................................................................................... 4
Mercury and Air Toxics Standards (MATS) ...................................................................... 4
Cross-State Air Pollution Rule (CSAPR) ........................................................................... 4
National Ambient Air Quality Standards (NAAQS) .......................................................... 5
Carbon Dioxide (CO2) ........................................................................................................ 5
Other Rules ......................................................................................................................... 5
3. Methodology ...................................................................................................................... 5
4. RFF’s Haiku Electricity Market Model ........................................................................ 12
5. Results .............................................................................................................................. 15
Coal Capacity .................................................................................................................... 15
Coal and Natural Gas Generation ..................................................................................... 20
Pollution Control Retrofits................................................................................................ 24
Emissions .......................................................................................................................... 26
6. Conclusion ......................................................................................................................... 27
References .............................................................................................................................. 31
Resources for the Future
Beasley et al.
Mercury and Air Toxics Standards Analysis Deconstructed:
Changing Assumptions, Changing Results
Blair Beasley, Matt Woerman, Anthony Paul, Dallas Burtraw, and Karen Palmer
1. Introduction
In 2011, the US Environmental Protection Agency (EPA) finalized two landmark
environmental regulations aimed at curbing air pollution from the electricity sector. The CrossState Air Pollution Rule (CSAPR) would have reduced sulfur dioxide (SO2) and nitrogen oxide
(NOx) emissions that are crossing state lines and contributing to pollution problems downwind.
The Mercury and Air Toxics Standards (MATS) is intended to reduce the emissions of hazardous
air pollutants, such as mercury and acid gases, from coal- and oil-fired power plants. Both rules
face legal challenges in the courts. In August 2012, the US Court of Appeals for the DC Circuit
threw out CSAPR, remanding the rule to EPA to try again. As a result, CSAPR’s predecessor,
the Clean Air Interstate Rule (CAIR), remains in effect. MATS also faces legal challenge;
however, the DC Circuit Court has agreed to take no action on a pending lawsuit while EPA
reconsiders the rule’s requirements for new power plants.
If CSAPR had remained in effect, EPA projected that the rule would have decreased SO2
emissions in the affected region in 2014 by 6.4 million tons per year compared to 2005 emissions
levels, to an expected level of 2.4 million tons. The rule would have decreased NOx emissions by
1.4 million tons per year, to an expected level of 1.2 million tons (2011d). In contrast, CAIR
achieves most of these reductions, although the levels vary among the states. If CAIR remains in
effect, it is expected to achieve emissions levels of 2.5 million tons of SO2 and 1.3 million tons of
NOx in affected states by 2015 (EPA 2012a). MATS, meanwhile, is estimated to reduce mercury
emissions from coal-fired power plants by 90 percent and reduce acid gas emissions by 88
percent. The rule is also projected to reduce SO2 emissions from power plants by 41 percent
beyond what was projected from CSAPR. These reductions come at a cost to the electricity
industry. EPA estimates that MATS will cost $9.6 billion a year— one of the most expensive
single regulations in the nation’s history (EPA 2011a). CSAPR was expected to cost $800

Beasley is a research assistant, Woerman is a research associate, Paul is a fellow in the Center for Climate and
Electricity Policy, Burtraw is the Darius Gaskins Senior Fellow, and Palmer is a senior fellow and research director
at Resources for the Future. The authors wish to thank the Bipartisan Policy Center for funding as well participants
of a workshop hosted by Resources for the Future and sponsored by the Electric Power Research Institute that took
place July 19, 2012. Direct correspondence to burtraw@rff.org and palmer@rff.org.
1
Resources for the Future
Beasley et al.
million in 2014, along with $1.6 billion a year in capital investments that were started under
CAIR (EPA 2011d). The costs are almost entirely the result of new pollution control investments
at coal-fired power plants.
The overall cost and the timing of the regulations have sparked debate within the
electricity industry about the impacts of the rules. There is particular concern that the new
requirements could cause reliability problems, including electricity supply disruptions, shortages
of generation capacity, or problems meeting reliability standards. This has led to some political
backlash, with various groups calling EPA’s suite of new air pollution rules, including some
rules that are anticipated but not yet proposed, a “regulatory train wreck” (Congressional
Research Service 2011). Others, however, have found that the focus on the reliability of supply is
unwarranted, with widespread supply disruptions resulting from CSPAR and MATS unlikely.
Several of those studies are analyzed in this paper, including work by the Bipartisan Policy
Center (Macedonia and Kelly 2012) and Resources for the Future (RFF) (Burtraw et al. 2012a).
However, the rules are not without impact. As shown in the RFF paper, MATS is likely to
impact reliability in terms of changes to current electricity prices, industry revenues and costs,
and profits.
Complicating the reliability debate is the impact of low natural gas prices and electricity
demand on the electricity sector. Burtraw et al. (2012b) found that changes in relative fuel prices
and electricity demand within the electricity sector that would have occurred in the absence of
EPA regulation are expected to have a much larger effect on changes in capacity and generation
mix than will MATS or CSAPR. Since the beginning of 2010, natural gas prices have fallen
considerably, driven by the new shale gas boom. In addition, electricity demand growth has
slowed in the wake of the economic downturn and increased energy efficiency. The lower gas
prices are expected to drive electricity generation away from coal-fired plants and toward natural
gas, while the reduced demand is associated with a reduction in the growth of electricity
generation. A comparison between a baseline scenario that uses 2009 natural gas supply and
electricity demand data and a baseline scenario that uses 2011 data highlights this trend. When
2011 data is used instead of 2009 data, coal generation falls by 14.7 percent. By comparison, the
addition of MATS to the 2011 baseline scenario reduced coal generation by only 1.5 percent.
All of this change in the industry and associated uncertainty have led many to consult
predictions from economic models that explore the impacts of EPA’s proposed rules. Several
studies have used economic models to attempt to quantify and explain the potential effects of
CSAPR and MATS, but these studies typically come to quite different conclusions. One reason
is that the analyses were done at different points in time with different information. Moreover,
the studies made different assumptions and used different models to reach their conclusions. It is
2
Resources for the Future
Beasley et al.
challenging to extract the effects of the regulations from the important differences in the
modeling assumptions used and the policies modeled. All of these variables contribute to a
general confusion about the likely impacts of the regulations.
This paper provides a review of several key analyses and attempts a reconciliation of the
results by deconstructing the differing assumptions within a common modeling platform. We
focus primarily on analyses of MATS, since CSAPR has been remanded to EPA. However,
several of the studies include CSAPR in their baselines or model CSAPR alongside MATS, and
we also investigate some of these scenarios. We conduct the reconciliation using RFF’s Haiku
electricity market model. This eliminates variability among the modeling platforms, one of the
key sources of variation among the studies. We then vary the assumptions and scenarios within
Haiku to represent choices made by other modeling teams.
In brief, our finding is that models that most closely mirror the requirements of the final
MATS rule and portray MATS without the confounding influence of other proposed EPA
regulations find the impact of the rule to be less severe. In particular, the way researchers
modeled the acid gas requirements under MATS had a large impact on the forecasted amount of
coal-fired capacity and generation going forward, as well as the pollution controls that would be
installed at these units and their emissions of acid gases. Those studies that allowed plants to
meet either a SO2 or a hydrochloric acid (HCl) emissions standard, as laid out in the final MATS
rule, tended to find less of an impact on coal-fired plants than those studies that forced plants to
strictly comply with the SO2 standard. As described below, the rule measures compliance for
some substances by proxy and provides flexibility for a facility in choosing among alternative
compliance strategies. The flexibility to choose which standard to meet, as set forth in the rule,
appears to lower costs importantly and reduces the impact on the coal fleet. Those studies that
forced plants to comply specifically with the SO2 standard also tended to find greater use of SO2
controls and thus greater reductions of SO2 emissions. In addition, the studies that introduced a
price on carbon dioxide (CO2) emissions or required selective catalytic reduction controls (SCR)
retrofits in anticipation of a future tightening of the NOx National Ambient Air Quality Standards
(NAAQS) saw relatively large reductions in predicted coal capacity and generation and
reductions in CO2 and NOx emissions, respectively, compared to studies that did not model these
future regulations.
3
Resources for the Future
Beasley et al.
2. EPA Policies That Are Modeled
Mercury and Air Toxics Standards (MATS)
EPA finalized MATS in December 2011. The rule regulates emissions of heavy metals
and acid gases from new and existing coal- and oil-fired power plants. This includes emissions of
mercury, arsenic, chromium, nickel, HCl, and hydrofluoric acid. The rule sets numerical
emissions limits for the pollutants and allows for some alternative compliance measures. For
example, under MATS, all coal-fired units must meet emissions limits for mercury, particulate
matter, and HCl. However, in place of the HCl standard, coal generators can opt to meet an
alternative SO2 limit.
To meet these standards, power plants can use a range of pollution control technologies,
such as wet and dry scrubbers, dry sorbent injection (DSI) systems, activated carbon injection
systems, fabric filters, and electrostatic precipitators. MATS requires plants to come into
compliance by 2015. However, EPA has said that it will allow state permitting authorities to give
an extra year to most plants to complete the retrofits if needed— pushing the compliance
deadline out to 2016 (EPA 2011c). In addition, EPA released a memo stating that plants that play
an important reliability role and are unable to comply by 2016 can apply for an additional year to
attain compliance (EPA 2011b). The rule faces legal challenges. However, the DC Circuit court
has agreed to take no action on the pending lawsuit while the agency reconsiders the rule’s
requirements for new power plants. EPA issued a proposed rule addressing emissions limits at
new power plants in November 2012. The agency does not expect the proposed rule to
significantly alter the costs, emissions reductions, or health benefits of MATS (EPA 2012b).
Cross-State Air Pollution Rule (CSAPR)
EPA finalized CSAPR in July 2011. The rule was written to replace the 2005 emissions
rule, CAIR, which was vacated by the DC Circuit Court in December 2008 and remanded to the
agency (EPA 2011d). However, in August 2012, the courts overturned CSAPR. As a result,
CAIR remains in effect.
CSAPR was intended to regulate NOx and SO2 air pollution that crosses state lines and
prevents downwind states in the eastern half of the country from meeting EPA standards for
ground-level ozone and fine particulate pollution. CSAPR covered a similar, but not identical,
group of states to those covered under CAIR. The first phase of CSAPR compliance would have
begun in January 2012, with a second phase of SO2 reductions beginning in 2014. Power plants
subject to the regulation were allowed to trade or bank emissions allowances. However,
4
Resources for the Future
Beasley et al.
individual states were subject to their own emissions caps, which could not be exceeded without
penalty (EPA 2011d).
National Ambient Air Quality Standards (NAAQS)
The Clean Air Act requires EPA to set NAAQS for six criteria pollutants: SO2, NOx,
ozone, particulate matter, carbon monoxide, and lead. EPA sets primary standards, which are
intended to protect public health, and secondary standards, which are intended to protect public
welfare. The law requires the agency to review, and if appropriate, revise the requirements every
five years. The NAAQS for SO2 and NOx were last updated in 2010 (EPA 2012c). Some
researchers have begun to model a tightening of the NOx NAAQS by requiring all coal plants to
install SCR units in some future year, while others have begun to model a tightening of the SO2
NAAQS by requiring all coal plants to meet an SO2 emissions rate standard more stringent than
that introduced by MATS.
Carbon Dioxide (CO2)
There has been a lot of debate in recent years about how to best limit CO2 emissions from
new and existing power plants. Due to uncertainty about the final regulatory or legislative
outcome, some modeling teams used a tax on CO2 as a proxy for a future CO2 emissions
standard. The size of the tax proxy varies from model to model. The models covered in this
review have assumed a CO2 tax that starts between $10 and $25 per ton and grows over time.
Other Rules
Several other environmental regulations relevant to the electricity sector have been
announced recently or are anticipated in the coming years. These include the Clean Water Act
316(b) rule, the Coal Combustion Residual rule, and Regional Haze rules. Although some
modelers have begun to include these rules in their analyses, they are not the focus of this study
and will not be discussed in detail.
3. Methodology
This paper compares results from nine studies that modeled the impacts of MATS and
attempts to reconcile differences in the study findings by conducting simulations of the MATS
rule using a common modeling platform. The simulations are run in Haiku and are intended to
explore the influence of key assumptions made by the studies on several key outcomes, such as
capacity mix, generation mix, emissions of relevant pollutants, and installations of pollution
control technologies.
5
Resources for the Future
Beasley et al.
The following lists the assumptions explored as well as the associated scenario number as
described in Table 1: a baseline against which changes are compared (1); whether the models
analyzed MATS only (2) or CSAPR alongside MATS (3); whether the models require plants to
comply with the acid gas standard under MATS by meeting the SO2 standard (4 and 9), the HCl
standard (5), or either standard (2 and 3); whether the models allowed DSI (2 and 3) or restricted
its installation as a compliance technology under MATS (6 and 9); whether the models required
all coal plants to install SCR units to simulate compliance with a future NOx NAAQS (7); or
whether models imposed a price on CO2 to simulate a future greenhouse gas policy (8). Table 1
summarizes the scenarios modeled in Haiku that simulate these assumptions.
Table 1. Summary of Scenarios Modeled in Haiku
Scenario name
1. Baseline
Policy modeled
This scenario includes CAIR and other existing state and federal
policies. It does not include CSAPR or MATS.
2. MATS
This scenario includes MATS as well as CAIR and other existing
state and federal policies.
3. MATS/CSAPR
4. MATS SO2
5. MATS HCl
This scenario includes MATS and CSAPR (instead of CAIR).
This scenario is identical to the MATS scenario, but with all plants
meeting the MATS SO2 standard for acid gases.
This scenario is identical to the MATS scenario, but with all plants
meeting the MATS HCl standard for acid gases.
6. MATS no DSI
This scenario is identical to the MATS scenario, but with no option
for DSI.
7. MATS SCR
This scenario is identical to the MATS scenario, but with all coal
plants required to install SCR to meet a future NOx NAAQS.
This scenario is identical to the MATS scenario, but with all fossil
8. MATS CO2
plants also facing a CO2 tax that begins in 2016 and increases
annually by 5 percent, hitting $10/ton in 2020.
9. MATS SO2 no DSI
This scenario is identical to the MATS SO2 scenario, but with no
option for DSI.
6
Resources for the Future
Beasley et al.
The studies reviewed in this paper are all national in scope and use economic models to
make predictions about the impacts of MATS on the electricity sector. Regional studies and
those that rely on qualitative analysis were not included.1 Table 2 summarizes the studies
analyzed in this report and provides an overview of some of the relevant assumptions made by
the modeling teams. Some studies modeled multiple scenarios. The table below only describes
the assumptions of the scenarios reviewed in this report.
1
For a more complete list of studies and additional analysis, see Beasley and Morris (2012) .
7
Resources for the Future
Beasley et al.
Table 2. Summary of Studies Reviewed
Organization,
authors, and
date of
publication
Bipartisan Policy
Center (BPC),
Macedonia and
Kelly, 2012 (July)
Charles River
Associates
(CRA),
Shavel and
Gibbs, 2012
(July)
Publication
title
BPC
Modeling
Results:
Projected
Impact of
Changing
Conditions
on the
Power
Sector
Electric
Sector
Impacts:
CSAPR,
MATS, etc.
Policies
analyzed
CSAPR,
MATS
CSAPR,
MATS,
proposed
CWA
316(b), $10
per metric
ton CO2
policy, NOx
policy
requiring
SCRs
Policies in
the
baseline
Existing
state and
federal
regulations
as of
summer
2011, such
as CAIR
CSAPR
Natural gas
assumptions
Treatment
of acid gases
under MATS
Limits on
installing
DSI
Coal
capacity
predicted
to retire in
the
baseline
Annual
Energy
Outlook
(AEO) 2012
early release
Model SO2
and HCl
compliance
options,
including HCl
compliance
using lowchlorine coal
with a DSI
All units
can install
DSI
40 GW by
2016
AEO 2012
and futures
through
2014
Model SO2
compliance
option, not
HCl. Most
plants can
comply with
wet
scrubbers.
Dry
scrubbers
are not
modeled.
Not all plants
can choose
to install DSI.
Only units
that burn
western
coal, are
smaller
than 200
MW, and
are not
utilityowned
can install
DSI
8
23.2 GW by
2020
Year in
which key
rules are
assumed
to take
effect
CSAPR
(2013,
2014),
MATS
(retrofits
of ACI, DSI
and ESP
upgrades,
2015;
scrubber
and fabric
filters,
2016)
CSAPR
(2013,
2014),
MATS
(2015 or
2017),
CWA 316
(b) (2020),
CO2 (2020)
Model
Closest
match to
Haikumodeled
scenario
IPM
MATS/
CSAPR
NEEM
MATS,
MATS SCR,
MATS CO2
Resources for the Future
Department of
Energy (DOE),
DOE, 2011
(December)
Edison Electric
Institute (EEI),
Fine, Fitzgerald,
and Ingram,
2011 (January)
Beasley et al.
Resource
Adequacy
Implications
of
Forthcoming
EPA Air
Quality
Regulations
CSAPR,
MATS
Potential
Impacts of
Environmental
Regulation
on the US
Generation
Fleet
Proposed
CSAPR,
MATS
(written
prior to
proposal),
CCR, CWA
316(b)
(written
prior to
proposal),
CO2
regulation,
NOx policy
requiring
SCRs
Existing
state and
federal
regulations,
such as
CAIR
Existing
state and
federal
regulations,
such as
CAIR and
state
mercury
standards
AEO 2011
Prices
constructed
from EPA
v4.10 supply
curves for
2015 and
2020
Require all
coal units to
install wet
scrubbers in
its stringent
scenario
Require all
coal plants
to install a
scrubber
9
Does not
allow DSI
in its
stringent
scenario
Does not
allow
8 GW by
2015
22 GW in
2015, 25
GW in 2020
CSAPR
(2012,
2014),
MATS
(2015)
MATS
(2015),
CWA
316(b)
(2022,
2027), CO2
(2017),
CCR
(Subtitle D)
(2017),
CSAPR
(2017,
2018)
PINEMS
MATS/
CSAPR,
MATS no DSI
IPM
MATS/
CSAPR,
MATS SCR
(primary
scenario);
MATS CO2
(alternative
CO2 policy
case)
Resources for the Future
Environmental
Protection
Agency (EPA),
EPA, 2011
(December)
Electric Power
Research
Institute (EPRI),
EPRI, 2012
(October)
Regulatory
Impact
Analysis for
the Final
Mercury and
Air Toxics
Standards
Prism 2.0:
The Value of
Innovation in
Environmental Controls
Beasley et al.
MATS
MATS,
CWA
316(b),
CCR, SO2,
NOx, O3,
and PM
NAAQS,
Regional
Haze
Existing
state and
federal
regulations,
CSAPR, no
state-level
mercury
standards
Existing
state and
federal
regulations,
such as
CSAPR
AEO 2011
AEO 2011
Model SO2
and HCl
compliance
options
Modeled SO2
compliance
option only
10
Units
must be at
least 25
MW and
burn lowsulfur coal
Not
allowed in
the
reference
case
7 GW by
2015
MATS
(2015)
IPM
MATS
5 GW by
2020
MATS
(2015),
CWA
316(b)
(2018),
CCR
(Subtitle D
by 2020),
SO2, NOx,
O3, and PM
NAAQS,
Regional
Haze
(2018)
USREGEN
MATS SO2
no DSI,
MATS SCR
Resources for the Future
KanORS/
DecisionWare
(KanORS),
Wright and
Kanudia, 2012
(July)
Analysis of
CSAPR and
MATS in
FACETS
NERA Economic
Consulting
(NERA),
Smith,
Bernstein,
Bloomberg,
Mankowski, and
Tuladhar, 2012
(March)
An Economic
Impact
Analysis of
EPA’s
Mercury and
Air Toxics
Standards
Rule
Resources for
the Future (RFF),
Burtraw, Palmer,
Paul, Beasley,
and Woerman,
2012 (May)
Reliability in
the
Electricity
Industry
under New
Environmental
Regulations
Beasley et al.
Existing
state and
federal
policies,
excluding
CAIR
MATS
Included
two
baselines,
one with
CSAPR and
one with
CAIR
CSAPR,
MATS
Existing
state and
federal
policies,
including
CAIR
CSAPR,
MATS
AEO 2011
Modeled HCl
compliance
option only
Plants 25
MW or
larger can
install DSI
8.5 GW by
2020
To account
for
uncertainty
in shale gas’s
impact on
prices, NERA
includes
resource
supply
curves for
US gas;
NERA also
includes
supply and
demand
curves for
US imports
and exports.
Modeled HCl
compliance
option only.
Plants could
use DSI or
scrubbers to
meet the
standard.
Plants could
also choose
to burn lowchlorine
coal.
Units
smaller
than 300
MW that
burn subbituminous coal
can install
DSI
15 GW by
2015 (in
the
baseline
with CAIR)
AEO 2011
Model SO2
and HCl
compliance
options
Units
must be at
least 25
MW and
burn lowsulfur coal
11
13.5 GW
CSAPR
(2014),
MATS
(2017)
MATS
(2015)
CSAPR
(2012,
2014),
MATS
(2016)
FACETS
MATS HCl
NewER
A
MATS HCl,
MATS no DSI
Haiku
MATS,
MATS/
CSAPR
Resources for the Future
Beasley et al.
4. RFF’s Haiku Electricity Market Model2
The Haiku model solves for investment and operation of the electricity system in 22
linked regions in the continental United States, starting in 2013 and solving out to the year 2035.
Each simulation year is represented by three seasons (spring and fall are combined) and four
times of day. For each time block, demand is modeled for three customer classes (residential,
industrial, and commercial) in a partial adjustment framework that captures the dynamics of the
long-run demand responses to short-run price changes. Supply is represented using 58 model
plants in each region. Thirty-nine of the model plants aggregate existing capacity according to
technology and fuel source from the complete set of commercial electricity generation plants in
the country. The remaining 19 model plants represent potential new capacity investments, again
differentiated by technology and fuel source. Each coal model plant has a range of capacities at
various heat rates, representing the range of average heat rates at the underlying constituent
plants.
Operation of the electricity system (generator dispatch) in the model is based on the
minimization of short-run variable costs of generation, and a reserve margin is enforced based on
margins obtained by the Annual Energy Outlook (AEO) for 2011 (EIA 2011). Fuel prices are
benchmarked to the AEO 2011 forecasts for both level and supply elasticity. Coal is
differentiated along several dimensions, including fuel quality and content and location of
supply, and both coal and natural gas prices are differentiated by point of delivery. The price of
biomass fuel also varies by region, depending on the mix of biomass types available and delivery
costs. All fuels are modeled with price-responsive supply curves. Prices for nuclear fuel and oil,
as well as the prices of capital and labor, are held constant.
Investment in new generation capacity and the retirement of existing facilities are
determined endogenously for an intertemporally consistent (forward-looking) equilibrium, based
on the capacity-related costs of providing service in the present and into the future (goingforward costs) and the discounted value of going-forward revenue streams. Existing coal-fired
facilities also have the opportunity to make endogenous investments to improve their efficiency.
Discounting for new capacity investments is based on an assumed real cost of capital of 7.5
percent. Investment and operation include pollution control decisions to comply with regulatory
constraints for SO2, NOx, mercury, HCl, and particulate matter (PM), including equilibria in
emissions allowance markets where relevant. Table 3 lists the pollution controls included in
2
For detailed documentation of Haiku, see Paul et al. (2009).
12
Resources for the Future
Beasley et al.
Haiku and the primary pollutants they control. All currently available generation technologies as
identified in AEO 2011 are represented in the model, as are integrated gasification combined
cycle coal plants with carbon capture and storage and natural gas combined cycle plants with
carbon capture and storage. Ultra-supercritical pulverized coal plants and carbon capture and
storage retrofits at existing facilities are not available in the model.
Table 3. Pollution Abatement Control Technologies Modeled in Haiku
Control technology
Acronym
Primary pollutants abated
SCR
NOx
SNCR
NOx
Wet flue gas desulfurization
Wet FGD
SO2 and HCl
Dry flue gas desulfurization
Dry FGD
SO2 and HCl
Dry sorbent injection
DSI
SO2 and HCl
Activated carbon injection
ACI
Mercury
Fabric filters
FF
Particulate matter
Electrostatic precipitator
ESP
Particulate matter
Selective catalytic reduction
Selective noncatalytic reduction
Price formation is determined by cost-of-service regulation or by competition in different
regions corresponding to current regulatory practice. Electricity markets are assumed to maintain
their current regulatory status throughout the modeling horizon; that is, regions that have already
moved to competitive pricing continue that practice, and those that have not made that move
remain regulated.3 The retail price of electricity does not vary by time of day in any region,
though all customers in competitive regions face prices that vary from season to season.
An important part of the model is the requirement that each region have sufficient
capacity reserve to meet requirements drawn from the North American Electric Reliability
Corporation. The reserve price reflects the scarcity value of capacity and is set just high enough
to retain sufficient capacity to cover the required reserve margin in each time block. The model
does not model separate markets for spinning reserves and capacity reserves. Instead, the fraction
of reserve services provided by steam generators is constrained to be no greater than 50 percent
of the total reserve requirement in each time block.
Table 4 outlines some key differences between assumptions used in the studies reviewed
for this paper and the assumptions made in Haiku.
3
There is currently little momentum in any part of the country for electricity market regulatory restructuring. Some
of the regions that have already implemented competitive markets are considering reregulating, and those that never
instituted these markets are no longer considering doing so.
13
Resources for the Future
Beasley et al.
Table 4. Differences between Study Assumptions and Scenarios as Modeled in Haiku
Sponsoring organization,
authors, and publication date
Bipartisan Policy Center (BPC),
Macedonia and Kelly, 2012
Charles River Associates (CRA),
Shavel and Gibbs, 2012
Department of Energy (DOE), DOE,
2011
Edison Electric Institute (EEI), Fine,
Fitzgerald, and Ingram, 2011
Environmental Protection Agency
(EPA), EPA, 2011
Electric Power Research Institute
(EPRI), EPRI, 2012
KanORS/DecisionWare (KanORS),
Wright and Kanudia, 2012
NERA Economic Consulting (NERA),
Smith, Bernstein, Bloomberg,
Mankowski, and Tuladhar, 2012
Resources for the Future (RFF),
Burtraw, Palmer, Paul, Beasley, and
Woerman, 2012
Key differences from scenarios as modeled in Haiku
BPC uses AEO 2012 early release electricity demand growth and natural gas forecasts, while Haiku uses AEO 2011. BPC
requires all plants that use low-chlorine coal to comply with the MATS acid gas standard to install DSI (although the
plant does not have to pay operating costs for the control). Haiku does not have this requirement.
CRA includes CSAPR in its baseline instead of CAIR. In its scenarios with a CO 2 price and SCR requirement, it also models
the Clean Water Act 316(b) rule, unlike Haiku. MATS compliance begins in 2015 or 2017; compliance begins in 2016 in
Haiku. CRA doesn’t allow dry scrubbers and significantly restricts DSI in its MATS modeling. It also requires all units to
install ACI to comply with MATS and doesn’t allow the combination of wet scrubbers and SCR to be used as a
compliance strategy. These options are allowed in Haiku. CRA uses AEO 2012 natural gas prices instead of AEO 2011.
DOE requires coal units to install wet scrubbers and fabric filters by 2015 to comply with MATS under its stringent
policy scenario. Plants cannot use DSI. DOE also models CSAPR SO2 regions 1 and 2 as one trading region. All new
investments in pollution controls must be paid back within 10 years. Haiku does not impose these restrictions.
In its primary air case, EEI requires every coal plant to install a scrubber, ACI, and fabric filter to comply with MATS by
2015. DSI is not allowed. In Haiku, all pollution control investments are selected endogenously, with the exception of
particulate matter controls. EEI also requires all coal plants in the eastern US to install SCRs by 2018. Haiku’s SCR case
requires SCR installation beginning in 2020. EEI models its primary air case along with the Coal Combustion Residual
rule and the Clean Water Act 316(b) rule. Haiku does not model these. The alternative CO2 case requires carbon
capture and sequestration beginning in 2020. Haiku does not impose this requirement. EEI imposes a carbon tax of $10
per ton in 2017, escalating at 5%. This is slightly different than in Haiku.
EPA includes CSAPR in the baseline instead of CAIR. Compliance with MATS begins in 2015 instead of 2016.
EPRI includes CSAPR in the baseline instead of CAIR. It also requires all plants to comply with the MATS SO 2 standard.
Plants can comply with the HCl or SO2 standard in Haiku. EPRI includes more policies than Haiku models, including the
Clean Water Act 316(b) rule and the Coal Combustion Residual rule. EPRI requires SCR units at all coal plants beginning
in 2018 instead of 2020. In its reference case, EPRI does not allow for DSI as a compliance option for MATS.
The KanORS model begins MATS compliance in 2017 instead of 2016. It requires all plants to comply with the MATS HCl
standard, whereas plants can comply with the HCl or SO2 standard in Haiku. Plants cannot retrofit with dry scrubbers.
NERA begins MATS compliance in 2015 instead of 2016. It also requires all plants to comply with the MATS HCl
standard, whereas plants can comply with the HCl or SO2 standard in Haiku. NERA limits DSI installations to units that
are smaller than 300 MW; Haiku does not impose this limit.
RFF uses the same MATS and MATS/CSAPR scenarios as those included in the Haiku reconciliation scenarios.
14
Resources for the Future
Beasley et al.
5. Results
The discussion of results focuses on several important electricity market outcomes,
including capacity and generation mix, pollution controls, and emissions, which constitute the
primary metrics examined in the different studies. For each result, we include a table that
compares results from the scenarios modeled in Haiku4 and the other studies. The sections on
capacity and generation also include a figure that displays the results graphically. A number of
the studies focus exclusively on capacity and pollution control changes, so they are not included
in the other results tables.
Some of the differences among the studies stem from variations in the studies’ baselines.
A summary of these differences is included in Table 2. For example, some studies, such as those
by BPC and CRA, relied on data about the electricity sector from the 2012 version of the AEO,
whereas the other studies used the 2011 version. There are some noteworthy differences between
the two AEO versions, including differences in natural gas prices. Studies that relied on 2011
data assumed higher natural gas prices than studies that relied on AEO 2012 data, which, as
noted, have the potential to drive out more coal capacity that either CSAPR or MATS. It is less
important, but also noteworthy, that studies differed in whether CSAPR or CAIR was included in
the baseline. These differences contributed to the varying estimates of coal retirements predicted
to occur in the baseline. Although differences in the baselines are important, we do not analyze
them in the discussion below.
Coal Capacity
The first set of results focuses on the effects of different scenarios on coal-fired capacity
retirements. The results are displayed in Table 5 as well as graphically in Figure 1. Table 5
shows coal capacity retirements in the baseline and incremental coal retirements under each of
the policy scenarios. Results for the various scenarios modeled in Haiku are compared to results
from the other modelers. The rows that report results from other models also identify the
scenario modeled in Haiku that is the closest match. Some scenarios match loosely with multiple
scenarios modeled in Haiku. For example, DOE models both CSPAR and MATS, which aligns
with the MATS/CSAPR scenario in Haiku. But DOE also restricts DSI and requires SO2
compliance, which aligns with the MATS SO2 no DSI scenario in Haiku. Therefore, both Haiku-
4
Results from Haiku are presented for the year 2016, but results for 2020 are very similar except where noted.
15
Resources for the Future
Beasley et al.
modeled scenarios are listed in the “Closest match to Haiku-modeled scenario” column in the
table below.
Table 5. Coal Capacity
Study
Haiku reconciliation
scenarios
BPC
CRA
DOE
EEI
EPA
Closest match to
Haiku-modeled
scenario
MATS
MATS/CSAPR
MATS SO2
MATS HCl
MATS no DSI
MATS SCR
MATS CO2
MATS SO2 no DSI
MATS/CSAPR
MATS/CSAPR & MATS
SO2 no DSI
MATS/CSAPR, MATS
CO2, & MATS SO2 no
DSI
MATS/CSAPR, MATS
CO2, MATS SCR, &
MATS SO2 no DSI
MATS/CSAPR & MATS
SO2 no DSI
MATS/CSAPR, MATS
SCR, & MATS SO2 no
DSI
MATS/CSAPR, MATS
SCR, MATS CO2, &
MATS SO2 no DSI
MATS/CSAPR
Year of
reported
results
2016
2016
Incremental coal
capacity predicted to
retire in policy scenarios
(GW)
3.8
3.6
38.4
3.2
6.7
15.0
26.7
39.9
16
Coal capacity
predicted to retire
in the baseline
(GW)
13.5
40
a
12.3
b
2020
84.5
23.2
c
120.9
2015
21
8
d
25.6
2020
25
e
48.9
2015
4.7
EPRI
MATS SO2 no DSI &
MATS SCR
2020
57-85
KanORS/DecisionWare
MATS HCl
2019
50
7
f
5
8.5
MATS/CSAPR & MATS
2015
23
15
HCl
MATS
3.8
RFF
2016
13.5
MATS/CSAPR
3.6
a
This reflects a scenario with MATS with a 2015 implementation date and CSAPR.
b
This reflects a scenario with MATS, CSAPR, a CO2 price, and the Clean Water Act 316(b) rule.
c
This reflects a scenario with MATS, CSAPR, a SCR requirement, a CO2 price, and Clean Water Act 316(b) rule.
d
This reflects a scenario with MATS, CSAPR, a SCR requirement, Clean Water Act 316(b) rule, and Coal
Combustion Residual rule.
e
This reflects a scenario with MATS, CSAPR, a SCR requirement, a CO2 price, Clean Water Act 316(b) rule, and
Coal Combustion Residual rule.
f
This reflects a scenario with MATS, future NAAQS requiring SCR and a more stringent SO 2 standard, Clean Water
Act 316(b) rule, Coal Combustion Residual rule, and future Regional Haze rules.
NERA
16
Resources for the Future
Beasley et al.
The results are also displayed graphically in Figure 1. The bars represent coal retirements
projected in each of the studies reviewed. The overlapping shapes represent coal retirements in
the scenarios modeled in Haiku that most closely match the original studies. For example, the bar
labeled BPC represents the 56 GW of coal capacity predicted to retire in the BPC study. This
includes capacity that is predicted to retire in the baseline and in the policy scenario. The
overlapping gray square represents the 17.1 GW of coal capacity predicted to retire in the
baseline and MATS/CSAPR scenario modeled in Haiku, which is the scenario that most closely
matches the assumptions made by BPC. Studies are listed more than once if multiple scenarios
are analyzed. Bars with multiple overlapping shapes represent original studies that combined
multiple policies and that map to more than one scenario modeled in Haiku. As discussed below,
when more than one policy is modeled, its impacts are not necessarily additive. However, as seen
in Figure 1, the additional costs of complying with multiple policies do have an impact on
affected units, with more policy requirements or limitations in the modeling associated with
more coal capacity retirements.
Figure 1. Projected Coal Capacity Retirements (GW) in Original Models and
Corresponding Scenarios Modeled in Haiku
160
140
Retirements in Original Studies
120
Haiku Reconciliation Scenarios:
MATS/CSAPR
100
MATS
80
MATS SO2 no DSI
60
MATS CO2
40
MATS SCR
20
MATS HCL
0
Findings Using the Haiku Model
The scenarios modeled in Haiku focus on different assumptions about compliance
options with MATS or single policy add-ons to MATS. Other than under the MATS SO2 no DSI
17
Resources for the Future
Beasley et al.
scenario, we did not use the Haiku model to evaluate combinations of assumptions or polices,
such as modeling MATS along with an SCR requirement and a CO2 price. Thus it is unclear
what level of coal retirements would result in the Haiku model under a combination of
assumptions. Two policies may negatively affect the same set of coal units, in which case coal
retirements would not increase beyond that seen in one of the policies alone, so the combination
would be less than additive. On the other hand, the combined effect of the two policies may
cause additional coal capacity to retire that would have survived under a single policy, in which
case the combination would be more than additive. It is not known which effect would be seen
under any given combination of assumptions, and it is likely that the effect would change
depending on which assumptions are combined. Consequently, the level of retirements observed
in Haiku should not be directly compared to those in other studies that combine multiple
assumptions in addition to MATS, but rather one may compare the general qualitative trends
observed in the different studies. For example, as seen in Figure 1, studies that include more
restrictions, and thus map to more of the scenarios modeled in Haiku, tend to predict more severe
impacts, such as more coal capacity retirements. Studies by CRA, DOE, EEI, EPRI, and NERA
all include multiple policy assumptions in addition to MATS.
Among the scenarios modeled in Haiku, the largest retirement of coal capacity occurs
when plants are forced to comply with the MATS SO2 standard without the option to meet the
alternative HCl standard; this occurs in two scenarios: the MATS SO2 and MATS SO2 no DSI
cases. A number of coal plants are able to meet the HCl standard by simply switching to a coal
type with a low HCl content. When this option is removed, as in the studies by CRA, DOE, EEI,
and EPRI, many of these plants must instead install new SO2 controls to meet the SO2 standard.
The costs of these additional controls are prohibitive for some plants leading this capacity to
retire. In both of these scenarios, natural gas capacity increases slightly to help fill in the void
from the retired coal plants.
The second-largest retirement of coal capacity occurs in MATS CO2, the scenario that
includes a tax on CO2. This outcome reflects the fact that coal is the most carbon-intensive fuel
modeled, and therefore, a tax on CO2 penalizes coal and incentivizes a switch to lower-carbon
sources of generation, including natural gas and renewables. The third-largest amount of coal
capacity retires under MATS SCR, where all coal units are required to install SCR for
compliance with a future tightening of the NOx NAAQS. In all of the other scenarios modeled,
such as MATS and MATS/CSAPR, substantially less coal capacity is expected to retire as a
result of the introduction of new environmental policies than are projected to retire in the
baseline due to changes in secular trends between 2009 and 2011. This trend was highlighted in
the RFF paper as an indication that MATS is unlikely to harm coal capacity to the point of
causing reliability problems.
18
Resources for the Future
Beasley et al.
Findings of the Original Studies
The other studies show a wide variation in projected coal retirements, but they generally
follow the same trends as the corresponding scenarios evaluated in the Haiku model. That is, the
most coal capacity retires when SO2 compliance is required for MATS, when a price is put on
CO2 emissions, or when additional controls are required to meet future regulations included in
the analysis. It is important to note that several of the studies, including CRA, DOE, and EEI,
require all or most of the coal fleet to install wet scrubbers. This can be viewed as not only
requiring these units to comply with the MATS SO2 standard, but actually forcing them to
overcomply with the standard, similar to the MATS SO2 no DSI scenario modeled using Haiku.
For example, CRA models three policy scenarios: one that includes MATS and CSAPR,
a second that adds the Clean Water Act 316(b) rule and a $10 per ton tax on CO2 beginning in
2020 that rises at 5 percent annually in real dollars, and a third that adds a SCR requirement to
simulate future NOx NAAQS. The last two simulations yield the largest amount of coal
retirements of all the studies examined, 84.5 GW and 120.9 GW, respectively. This can be
expected due to the compounding effect of requiring scrubbers for MATS compliance, imposing
a price on emissions of CO2, and requiring SCR for NAAQS compliance, which conforms to the
pattern of higher retirements with more regulations seen in the scenarios modeled in Haiku.
The EEI simulations also result in a substantial amount of coal retirements. Although this
study includes many modeling scenarios, we chose to highlight two. The first includes MATS,
CSAPR, a SCR requirement, the Clean Water Act 316(b) rule, and the Coal Combustion
Residual rule, while the second adds to this a CO2 price starting at $10/ton in 2017. As noted
above, all coal plants in the EEI simulations are required to install scrubbers for MATS
compliance. These two scenarios yield 25.6 GW and 48.9 GW of coal retirements, respectively.
These results also agree with the findings using Haiku that requiring compliance with the MATS
SO2 standard, requiring additional controls, or imposing a CO2 price would greatly increase the
retirements of coal units.
EPRI also finds large retirements of coal capacity by 2020, ranging from 57 GW to 85
GW. Its model includes MATS, the Clean Water Act 316(b) rule, and the Coal Combustion
Residual rule, and it imposes additional emissions limits to simulate future changes to SO2, NOx,
ozone, and particulate matter NAAQS and Regional Haze rules. Additionally, this model forces
coal plants to comply with the SO2 standard in MATS without the option to meet the alternative
HCl standard or install DSI. These results fit with the trends discussed previously that increase
the retirement of coal units.
The EPA, BPC, and RFF models result in some of the fewest coal retirements, 4.7 GW,
16 GW, and 3.6 to 3.8 GW, respectively. These are also the models with assumptions closest to
19
Resources for the Future
Beasley et al.
those in the MATS/CSAPR and MATS scenarios modeled in Haiku, with no additional policies
in place and no restrictions on how plants can comply with MATS. The findings from the EPA,
BPC, and RFF studies are again in agreement with the results observed in the scenarios modeled
in Haiku, where the fewest coal retirements occur when no restrictions are imposed on MATS
compliance and no future policies are simulated. The difference between modeling CSAPR, as in
the MATS/CSAPR scenario, and CAIR, as in the MATS scenario, has little impact on coal
retirements.
These results suggest that the wide variation in coal retirements among models and
scenarios can be largely attributed to the way MATS is represented by the different modelers as
well as the accompanying policies that are included in the analysis. One exception to these trends
found in Haiku is the KanORS/DecisionWare model, which results in 50 GW of coal retirements
despite modeling a scenario very similar to the Haiku MATS HCl scenario, which yielded very
little retirement. The cause of this discrepancy is unclear, but it is likely due to model-specific
characteristics.
Coal and Natural Gas Generation
The next set of results focuses on coal and natural gas generation. Table 6 presents
changes in each compared to the baseline scenario for Haiku results and for those scenarios from
the other modelers where these variables are reported.
The results are also displayed graphically in Figures 2 and 3. The graphs highlight that
generation, like capacity, is affected more when the scenarios deviate from the parameters of the
final version of MATS.
20
Resources for the Future
Beasley et al.
Table 6. Coal and Natural Gas Generation
Study
Haiku reconciliation
scenarios
BPC
CRA*
EEI
Closest match to
Haiku-modeled
scenario
MATS
MATS/CSAPR
MATS SO2
MATS HCl
MATS no DSI
MATS SCR
MATS CO2
MATS SO2 no DSI
MATS/CSAPR
MATS/CSAPR & MATS
SO2 no DSI
MATS/CSAPR, MATS
CO2, & MATS SO2 no
DSI
MATS/CSAPR, MATS
CO2, MATS SCR, &
MATS SO2 no DSI
MATS/CSAPR, MATS
SCR, & MATS SO2 no
DSI
Year
2016
2020
2020
Change in coal generation
from the baseline
-3.7%
-1.2%
-8.5%
-3.0%
-4.7%
-8.1%
-13.0%
-9.3%
-3%
Change in natural
gas generation
from the baseline
2.3%
-2.4%
6.8%
2.8%
3.1%
8.5%
13.1%
8.1%
5%
-4.3%
5.9%
-29%
39.7%
-41.2%
56.4%
-7.9%
12.9%
2020
MATS/CSAPR, MATS
SCR, MATS CO2, &
-16.2%
19.4%
MATS SO2 no DSI
EPA
MATS/CSAPR
2015
-1.3%
3.1%
KanORS/DecisionWare
MATS HCl
2019
-15.8%
14.2%
MATS
-3.7%
2.3%
RFF
2016
MATS/CSAPR
-1.2%
-2.4%
*The CRA report does not include the total baseline generation in TWh. The percentages above are based on a
Haiku-modeled scenario with CSAPR and no MATS in 2020, which closely matches the CRA baseline. The total
generation from the scenario modeled in Haiku is 21,930 TWh.
21
Resources for the Future
Beasley et al.
Figure 2. Change in Coal Generation from the Baseline
30%
20%
Change in Coal Generation in Study
Scenarios
10%
Haiku Reconciliation Scenarios:
MATS/CSAPR Coal
MATS Coal
0%
MATS SO2 no DSI Coal
-10%
MATS CO2 Coal
-20%
MATS SCR Coal
-30%
MATS HCL Coal
-40%
-50%
Figure 3. Change in Natural Gas Generation from the Baseline
60.0%
50.0%
Change in Gas Generation in Study
Scenarios
40.0%
Haiku Reconciliation Scenarios:
MATS/CSAPR Gas
MATS Gas
30.0%
MATS SO2 no DSI Gas
20.0%
MATS CO2 Gas
MATS SCR Gas
10.0%
MATS HCL Gas
0.0%
-10.0%
22
Resources for the Future
Beasley et al.
Findings Using the Haiku Model
As seen in Table 6 and Figure 2, coal generation, like coal capacity, decreases most in the
scenarios modeled in Haiku where units must comply with the MATS SO2 standard, there is a
price on CO2, or SCR is required to simulate a future change in the NOx NAAQS. These
scenarios also yield a commensurate increase in natural gas generation to make up for the
reduced generation from coal.
The largest change to coal generation, a decrease of 13 percent, occurs in the MATS CO2
scenario, where the introduction of a price on CO2 provides a disincentive for generation from
coal units. The next largest changes, decreases of roughly 9 percent, result from the MATS SO2
no DSI and MATS SO2 scenarios, respectively, both of which require coal units to comply with
the MATS SO2 standard without the option to instead comply with the HCl standard. The MATS
SCR scenario results in coal generation decreasing by 8.1 percent. In the last three scenarios,
coal generation decreases because of coal capacity retirements, and the extent of the changes in
these two variables is roughly proportional.
Findings of the Original Studies
As in the Haiku-modeled scenarios, the studies that included a price on CO2, required
compliance with the MATS SO2 standard, and required SCR retrofits resulted in the largest
reductions in coal generation. The CRA study, for example, predicted a 41.2 percent decrease in
coal generation in its most stringent case. Much of this lost coal generation was replaced by
natural gas generation, which increased by 56.4 percent. EEI’s study also predicted large
decreases in coal capacity. In its most stringent case, coal generation decreases by 16.2 percent,
while natural gas increases by 19.4 percent. Natural gas also increases in the scenarios modeled
in Haiku to make up for lost coal generation. For example, in the MATS CO2 scenario, coal
generation decreases by 13 percent, while natural gas generation increases by 13.1 percent.
Conversely, the modeling by EPA, BPC, and RFF projects small changes in coal
generation. RFF predicts decreases of 3.7 to 1.2 percent, EPA predicts decreases of 1.3 percent,
and BPC predicts decreases of 3 percent. As mentioned above, these are the scenarios that most
closely duplicate the assumptions in the MATS/CSAPR scenario modeled in Haiku and in the
study by RFF, which yield very little change in the electricity generation mix. As with coal
retirements, this suggests that differences in coal generation can be attributed primarily to
differences in the assumptions used to model MATS and the inclusion of additional policies.
23
Resources for the Future
Beasley et al.
Pollution Control Retrofits
The third set of results is the change in the amount of capacity retrofitted with different
types of pollution controls. Aggregate results are reported in Table 7 for Haiku and for each of
the other models that report this information.
Table 7. Pollution Control Retrofits
Closest match
to Haikumodeled
scenario
Study
Haiku
reconciliation
scenarios
BPC
EEI
EPA
KanORS/
DecisionWare
NERA
RFF
MATS
MATS/CSAPR
MATS SO2
MATS HCl
MATS no DSI
MATS SCR
MATS CO2
MATS SO2 no
DSI
MATS/CSAPR
MATS/CSAPR,
MATS SCR, &
MATS SO2 no
DSI
MATS/CSAPR,
MATS SCR,
MATS CO2, &
MATS SO2 no
DSI
MATS/CSAPR
MATS HCl
MATS/CSAPR &
MATS HCl
MATS
MATS/CSAPR
Change in
wet FGD
from the
baseline
(GW)
10
20
29
12
16
-1
4
Change in
dry FGD
from the
baseline
(GW)
-1
-1
94
-1
30
-1
-4
Change in
DSI from
baseline
(GW)
Change in
SCR from
baseline
(GW)
Change in
ACI from
baseline
(GW)
Change in
FF from
baseline
(GW)
62
66
-4
62
-4
68
54
8
4
3
8
1
150*
0
139
134
105
137
135
117
124
53
53
36
53
52
51
43
30
92
-4
2
101
36
16
11
65
4
131
165
-4
~55
Not
modeled
88
~200
160
-5
~45
Not
modeled
69
~185
146
2015
-6
43
0
141
2017
77
22
Not
modeled
11
0
54
101
Not
reported
2015
1
47
22
1
71
124
2016
10
20
-1
-1
62
66
8
4
139
134
53
53
Year
2016
2016
2020
*This result is for 2020, the first year of the SCR requirement in Haiku modeling
Findings Using the Haiku Model
In the scenarios using the Haiku model, the modeling assumptions have an important
effect on the retrofit of SO2 and HCl controls. In many of the scenarios, approximately 60 GW of
DSI are installed to comply with the HCl or SO2 standard imposed by MATS. When units are
required to comply with the SO2 standard, however, DSI is not a sufficient control strategy and is
not used. Instead, approximately 120 GW of wet and dry FGD controls are installed, which
would raise the cost of compliance. Under the MATS no DSI scenario, which still allows for
compliance with the HCl standard but does not allow for the use of DSI, 46 GW of wet and dry
FGD are installed.
24
Resources for the Future
Beasley et al.
These changes in SO2 and HCl controls, or an SCR requirement, have a small effect on
the use of ACI to control mercury emissions. The combination of an FGD and SCR is able to
control some emissions of mercury, although the exact rate of emissions reductions depends on
unit-specific characteristics. Thus, installing FGD to comply with the SO2 or HCl standard in
MATS, or the requirement to install SCR to simulate future NOx NAAQS, may also bring the
unit into compliance with the MATS mercury standard. We find that the scenarios that require
SO2 compliance under MATS or require an SCR yield the lowest retrofits of ACI, approximately
105 GW and 117 GW, respectively. Most of the other scenarios result in roughly 135 GW of
new ACI.
The amount of fabric filter retrofits required for MATS compliance is directly related to
retirements of coal plants. When minimal coal units retire, 53 GW of fabric filter are installed.
Under the scenarios that result in larger retirements, however, this capacity falls to only 36 GW.
This suggests that the additional costs of these modeling assumptions, including the requirement
to comply with the MATS SO2 standard or imposing a price on CO2 emissions, fall primarily on
units that already have to bear the cost of retrofitting with fabric filters in order to comply with
the MATS rule.
The addition of SCR at coal units occurs in a substantial amount only under the scenario
where such retrofits are required. In the other modeling scenarios, there is little change in the
amount of SCR.
Findings of the Original Studies
In the other studies, ACI and fabric filters are retrofit in large quantities, as much as 200
GW of ACI in the EEI study and 165 GW of fabric filters in the BPC study. This conforms to the
trend seen in the modeling scenarios using Haiku, where, although modeling assumptions can
cause small changes in the amount of ACI and fabric filter retrofits, the overall trend is that large
capacities of both controls are required regardless of the scenario modeled.
Controls for SO2 and HCl, however, vary greatly among these different studies. In fact,
most studies project that the largest change will occur for a different control type. The EPA and
BPC studies find 43 GW and 65 GW of DSI, respectively; the KanORS/DecisionWare study
finds 77 GW of wet FGD; and the EEI and NERA studies find roughly 45 to 55 GW and 47 GW
of dry FGD, respectively. This also agrees with the patterns seen in the Haiku modeling, where
SO2 and HCl controls appear to be the most sensitive to modeling assumptions.
25
Resources for the Future
Beasley et al.
Emissions
The last modeling outcome that we compare is the effects of different scenarios on
emissions of a host of different gases, including SO2, NOx, CO2, and mercury. The results are
presented in Table 8.
Table 8. Pollution Emissions
Study
Haiku
reconciliation
scenarios
BPC
EPA
Change in SO2
emissions from
the baseline
(million tons)
Change in NOx
emissions from
the baseline
(million tons)
Change in CO2
emissions
from baseline
(million tons)
Change in
mercury
emissions
from baseline
(tons)
MATS
-1.3
-0.1
-61
-20.5
MATS/CSAPR
-1.4
0.2
-13
-20.4
MATS SO2
-3.0
-0.1
-123
-21.3
-1.3
-0.1
-45
-20.5
-1.5
-0.1
-73
-20.6
MATS SCR
-1.2
-0.6
-110
-19.9
MATS CO2
-1.5
-0.1
-202
-21.0
MATS SO2 no
DSI
-3.0
-0.1
-133
-21.3
Closest match
to Haikumodeled
scenario
b
EPRI
Year
MATS HCl
MATS no DSI
2016
a
MATS/CSAPR
2016
-2.0
-0.1
-71
-20.9
MATS/CSAPR
2015
-1.3
0.0
-15
-19.9
MATS SO2 no
DSI &
2020
-2.4
-0.8
-110
Not reported
2019
-5.7
-0.3
Not reported
-36.2
-1.3
-0.1
-61
-20.5
MATS SCR
KanORS/
DecisionWare
RFF
MATS HCl
MATS
2016
c
MATS/CSAPR
-1.4
0.2
-13
-20.4
This result is for 2020, the first year of the SCR requirement in Haiku modeling.
b
These results are for all EGUs, not just EGUs covered by the regulations.
c
Does not include CAIR in the baseline, so SO2 reductions appear much larger than in other models. If CSAPR is
included in the baseline, then SO2 reductions from MATS are comparable to those observed in the BPC and EPA
models.
a
MATS is intended to reduce emissions of mercury and SO2, and that is what we find in
the scenarios using the Haiku model. Mercury emissions are reduced by roughly 80 percent,
regardless of the assumptions used in the modeling scenario, and comparable reductions are
observed in the other studies. Emissions of SO2 are reduced most under the Haiku modeling of
the MATS policy that forces plants to comply with an SO2 standard rather than giving them the
26
Resources for the Future
Beasley et al.
option to instead comply with an HCl standard, with reductions of 3 M tons compared to at most
1.5 M tons in the other scenarios. This trend is also seen in the other studies. The EPRI study,
which requires compliance with the MATS SO2 standard, yields reductions of 2.4 M tons, which
is larger than the reductions observed in the BPC and EPA models.
The MATS rule, as well as the additional policies that some modelers included in their
analyses, also affects emissions of CO2 and NOx. In the Haiku modeling, CO2 reductions are
largest when a price is imposed on CO2, as would be expected. The next largest reductions of
CO2 emissions occur in those scenarios with the largest retirements of coal units and reductions
in coal generation. The remaining scenarios result in only small reductions of CO2 emissions.
This is also observed in the BPC and EPA models, which yield small reductions in CO2 under
assumptions that most closely replicate the assumptions of the MATS/CSAPR scenario in Haiku.
Emissions of NOx fall by the most in the MATS SCR scenario in Haiku, 0.6 M tons in 2020, the
first year of the SCR requirement. This trend is seen in the other studies as well. The EPRI study,
which also includes an SCR requirement, yields NOx reductions of 0.8 M tons, which is much
larger than the reductions observed in the other models.
6. Conclusion
Several recent studies have used economic simulation models to attempt to quantify and
explain the potential effects of EPA’s MATS rule. These studies, however, have tended to
produce inconsistent results, making it difficult to draw general conclusions about the effect the
rule will have on the electricity sector, especially the capacity of coal-fired units that will be
forced to retire as a consequence of the new regulation. We attempt to reconcile these differences
and extract the effect of MATS from the important differences in the modeling assumptions used
and the policies modeled.
We conclude that studies that most closely match the regulatory requirements as laid out
in the final MATS rule and do not include other proposed EPA regulations in their analyses find
MATS to have less severe impacts on the electricity market. This includes the studies that map
closest to the MATS and MATS/CSAPR scenarios modeled in Haiku, such as scenarios modeled
by BPC, EPA, and RFF. All of the other scenarios modeled in the studies we reviewed include at
least one additional policy requirement that differentiates it from the final MATS rule and makes
compliance more difficult for coal-fired units.
In particular, the way researchers modeled the acid gas requirements under MATS had a
substantial impact on the amount of coal-fired capacity projected to retire. Those studies that
allowed plants to meet either a SO2 or HCl emissions standard, as laid out in the final MATS
rule, tended to find less of an impact on coal-fired plants than those studies that forced plants to
27
Resources for the Future
Beasley et al.
comply with the SO2 standard. The inclusion of additional policies, such as a price on CO2 or a
requirement of SCR to simulate a future tightening of the NOx NAAQS, also cause more coalfired capacity to retire than under modeling of the MATS rule alone.
These same factors are also important for projections of the generation mix.
Consideration of the potential cascade of additional policies may have value, but unfortunately
the effect of the additional policies has been conflated with the effects of MATS per se, leading
to confusion, which this paper helps resolve. Coal-fired generation is most negatively affected
when a price is put on CO2, directly introducing a disincentive to generation from coal. Forcing
coal units to comply with the MATS SO2 standard or requiring SCR retrofits also decreases coal
generation because of the retirement of coal-fired capacity. In these cases, the reduced generation
from coal is made up almost entirely by higher natural gas generation.
Controls for SO2 and HCl emissions are the most sensitive to changes in modeling
assumptions. The modeling of the MATS acid gas requirements and the inclusion of additional
policies appear to affect the amount of SO2 and HCl controls required and what types of controls
are selected by coal-fired units. Conversely, retrofits of ACI to control mercury and fabric filters
to control particulate matter are required in large amounts for compliance with MATS, regardless
of the various scenarios modeled.
Due to changes in SO2 controls, emissions of SO2 are also highly dependent on modeling
assumptions. Forcing coal units to comply with the MATS SO2 standard approximately doubles
the reductions of SO2 emissions. Emissions of CO2 also vary with modeling scenarios, often
commensurate with the amount of coal-fired generation. Emissions of mercury, however, are
reduced by roughly 80 percent regardless of modeling assumptions or policies, and emissions of
NOx change substantially only when SCR is required for compliance with a future NOx NAAQS.
Although we have identified these general trends in studies of EPA’s MATS rule,
differences still remain among studies that cannot be easily explained. This suggests that the
effects of MATS are still uncertain. However, this reconciliation analysis narrows that
uncertainty significantly by identifying and explaining some of the largest differences among the
existing studies, making it easier to compare these studies and better understand how the MATS
rule will impact the electricity sector.
28
Resources for the Future
Beasley et al.
References
Beasley, B., Morris, D. 2012. Modeling the Electricity Sector: A Summary of Recent Analyses of
New EPA Regulations. Discussion Paper 12-52. Washington, DC: Resources for the
Future.
Burtraw, D., Palmer, K., Paul, A., Beasley, B., Woerman, M. 2012a. Reliability in the Electricity
Industry under New Environmental Regulations. Discussion Paper 12-18. Washington,
DC: Resources for the Future.
Burtraw, D., Palmer, K., Paul, A., Woerman, M. 2012b. Secular Trends, Environmental
Regulations and Electricity Markets. The Electricity Journal 25: 35-47.
Congressional Research Service. 2011. E.P.A.'s Regulation of Coal-Fired Power: Is a "Train
Wreck" Coming? Washington, DC.
http://www.lawandenvironment.com/uploads/file/CRS-EPA.pdf.
DOE (Department of Energy). 2011. Resource Adequacy Implications of Forthcoming EPA Air
Quality Regulations. Washington, DC : DOE.
EIA (Energy Information Administration). 2011. Annual Energy Outlook 2011. Washington DC.
EPA (Environmental Protection Agency). 2011a. Benefits and Costs of Cleaning Up Toxic Air
Pollution From Power Plants.
———. 2011b. The Environmental Protection Agency's Enforcement Response Policy for Use of
the Clean Air Act Section 113(a) Administrative Orders in Relation to Electric Reliability
and the Mercury Air Toxics Standard.
———. 2011c. Fact Sheet: Mercury and Air Toxics Standards for Power Plants.
———. 2011d. Fact Sheet: The Cross-State Air Pollution Rule: Reducing the Interstate
Transport of Fine Particulate Matter and Ozone.
———. 2011e. Regulatory Impact Analysis for the Final Mercury and Air Toxics Standards.
Washington, DC: EPA. http://www.epa.gov/ttn/ecas/regdata/RIAs/matsriafinal.pdf.
———. 2012a, July 31, 2012. "Clean Air Interstate Rule: Basic Information."
http://www.epa.gov/cair/basic.html (accessed March 28, 2013)
———. 2012b. Fact Sheet: Proposed Updates of the Limits for New Power Plants Under the
Mercury and Air Toxics Standards (MATS).
———. 2012c, "National Ambient Air Quality Standards (NAAQS)."
http://www.epa.gov/air/criteria.html (accessed 2012)
29
Resources for the Future
Beasley et al.
EPRI. 2012. Prism 2.0: The Value of Innovation in Environmental Controls. Palo Alto, CA:
EPRI.
Fine, S., Fitzgerald, S., Ingram, J. 2011. Potential Impacts of Environmental Regulation on the
US Generation Fleet. Washington, DC: Edison Electric Institute.
Macedonia, J., Kelly, C. 2012. BPC Modeling Results: Projected Impact of Changing Conditions
on the Power Sector. Washington, DC: Bipartisan Policy Center.
Paul, A., Burtraw, D., Palmer, K. 2009. Haiku Documentation: RFF's Electricity Market Model
version 2.0. Washington, DC: Resources for the Future.
Shavel, I., Gibbs, B. Slides Presented 2012. Electric Sector Impacts: CSAPR, MATS, etc.
Washington, DC: Charles River Associates.
Smith, A., Bernstein, P., Bloomberg, S., Mankowski, S., Tuladhar, S. 2012. An Economic Impact
Analysis of EPA's Mercury and Air Toxics Standards Rule. Washington, DC: NERA
Economic Consulting.
Wright, E., Kanudia, A. 2012. Analysis of CSAPR and MATS in FACETS.
KanORS/DecisionWare.
30